Relative contributions of low- and high-luminance components to material perception

Area of Science:

• Visual perception
• Material science
• Image processing

Background:

• Object material recognition relies on visual cues beyond specular highlights, including shading and texture.
• Lower luminance image components, like shading, are vital for material identification.
• Understanding the contribution of luminance variations to material perception is key.

Purpose of the Study:

• To psychophysically investigate the roles of low- and high-luminance image components in material perception.
• To determine how preserving different luminance ranges affects the perception of material properties.
• To explore the relationship between luminance statistics and material categorization.

Main Methods:

• Creation of two types of luminance-modulated images: low-luminance-preserved (LLP) and high-luminance-preserved (HLP).
• Psychophysical experiments where observers compared modified images to original photographs of stone, wood, leather, and fabric.
• Analysis of observer choices and correlation with luminance contrast and spatial-frequency sub-band statistics.

Main Results:

• Observers equally chose between LLP and HLP images, indicating a complex interplay of luminance components.
• HLP image choice ratios varied significantly across material samples and categories.
• A moderate correlation was found between HLP image choices and higher-spatial-frequency sub-band luminance statistics, particularly for leather and fabric.

Conclusions:

• Both lower- and higher-luminance image components significantly contribute to material perception, with their relative importance being material-dependent.
• While luminance statistics correlate with perception for materials like leather and fabric, more complex image features may be necessary for discerning stone and wood.
• The findings highlight the nuanced visual processing involved in material identification from images.